AI Engineer – ASR

Pratham International

. Lead development and deployment of speech recognition systems for automated reading assessment across multiple languages .

Posted 10/15/2026full-timeRemote • IndiaMid-LevelSeniorWebsite

Tech Stack

Tools & technologies

PythonPyTorch

About the role

Key responsibilities & impact

Lead development and deployment of speech recognition systems for automated reading assessment across multiple languages
Work across the full ASR pipeline, from data collection, annotation, data preparation, and model training to decoding, alignment, error analysis, and deployment
Provide technical inputs to speech data collection design across languages, dialects, age groups, reading levels, environments, devices, and target items
Define annotation guidelines and quality checks for transcripts, reading errors, disfluencies, background speech, and related phenomena
Identify data gaps and annotation inconsistencies through model error analysis
Recommend targeted data collection or re-annotation
Design annotation schemas and inter-annotator quality processes
Train, fine-tune, and evaluate ASR models for children’s read speech across languages and accents
Build pipelines for audio preprocessing, segmentation, normalization, transcription alignment, and model training
Adapt ASR architectures such as Whisper, wav2vec 2.0, Conformer, CTC/RNN-T, or similar models
Optimize models for low-bandwidth, store-and-forward, server-based, and potentially offline/mobile inference
Detect substitutions, omissions, insertions, repetitions, partial-word attempts, hesitations, and self-corrections
Implement forced alignment, phoneme-level alignment, or pronunciation-scoring approaches where required
Conduct systematic error analysis by language, reading level, age, accent, gender, recording conditions, and device
Collaborate with assessment experts and field teams to translate model outputs into reading accuracy, WCPM, fluency, and reading-level measures
Define and track ASR metrics including WER/CER, word-level accuracy, error-detection precision/recall, alignment accuracy, and reading-score agreement with human evaluators
Build reproducible experimentation and model-evaluation pipelines
Maintain clear technical documentation

Requirements

What you’ll need

A Degree in any Mathematics/Computer Science/Data Science or related STEM field
3-5 years of full-time experience; at least 2 working in an AI/ML role
Strong experience in automatic speech recognition, speech processing, or spoken-language technologies
Strong Python skills and experience with PyTorch or similar deep-learning frameworks
Experience with modern ASR architectures and toolkits such as Hugging Face Transformers, NeMo, Kaldi, or equivalent
Understanding of acoustic modelling
Understanding of CTC, transducer, and encoder-decoder architectures
Understanding of decoding and beam search
Understanding of language models
Understanding of forced alignment
Understanding of phoneme representations
Understanding of WER/CER and ASR error analysis
Experience preparing and working with large-scale audio and transcription datasets
Ability to diagnose model errors systematically rather than relying only on aggregate transcription accuracy

Benefits

Comp & perks

Flexible, remote-first work environment
Work from anywhere in India
Periodic travel to field locations
Autonomy to work from where most productive
Opportunity to see models in use and gather real-world feedback
Competitive remuneration benchmarked against industry standards and commensurate with experience
Impact at a global scale supporting learning journeys of millions of children
Opportunity to work on unique technical challenges in AI/ML, ASR, and low-bandwidth/offline environments
Autonomy to experiment with latest SOTA models
Work alongside a global, mission-driven team
Inclusive and collaborative environment
Equal opportunity workplace